NERO

The complex nature of mobile robot tasks leads to the necessity of systems with several coordinated robots (agents) working in cooperation. Many important societies related to robotics and automation (IEEE) or the European Commission agree with this idea. So, some directives of these institutions re fer to robotic elements connected to the communication nets or wireless nets including the robots themselves, the sensors distributed in the working place (static perception agents) exchanging and sharing information. This concept is extended to robot interactions with humans, with the sensors and with the environment. We propose this project which is very related with previous MCYT projects obtained by this research team, to continue working on robotics with three objectives: 1. Multi-robot cooperation techniques with task allocation based on optimization methods and on heuristics to work with heterogeneous robots in motion and in perception capabilities. Using the constraints imposed by the environment, the sensors and the communications, some cost functions for the operation strategies will be defined. The dynamic information of the environment will be used, taking into account the planned mission. Environmental and behaviour learning techniques will be developed for the topological map building, the localization of the robots and the environment understanding in terms of relevant features of the scene. 2. We will work in visual perception and visual control. The matching of features is one of the key cues of computer vision and it will be studied introducing invariant features or combining classical features with more modern invariant features. The constraints imposed by the geometry will be included for two or three images depending on the particular case. Multiple views will be extended to multiple robots and the intrinsic problem of camera calibration related to the use of different cameras will be considered. The interesting perception-action problem will be studied with information of the current robot or the information of other robots or even the information of other static perceptual agents to make the global system more efficient and more robust. 3. In the context of communications we will study the communication of the robots to each other and with the head agent. We will deal with ad-hoc mobile nets to work with real time traffic in such a way that no communication infrastructure will be needed. Fixed time protocols and performance analysis techniques in the net and in the robot tasks will be developed. On the other hand particular ideas on propagation of electromagnetic fields in dissipative media (ground, stones, iron..) and its application to different modulations and different frequencies will be studied.

Luis Riazuelo
Luis Riazuelo
Assistant professor

Related